Delta Sigma: An Inconvenient Truth

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by k4rstar, Jun 21, 2020.

  1. Raimei Templar

    Raimei Templar Friend

    Friend
    Joined:
    Dec 8, 2019
    Likes Received:
    430
    Trophy Points:
    63
    Location:
    Osaka
    Thanks for that, didnt realize it was sort of packaging it as PCM without actually converting anything.

    Another point I can add is that Sabre dacs actually cant play one of my DSD albums at all. I got Michael Jackson's Thriller from Acoustic Sounds hi res store a few years ago and sabre dacs actually straight up cant play it. It starts,stops click pops and generally sounds like pure distorted ass. Burr Brown based dacs play it without any issues at all and it sounds great. Something about the way Sabre does DSD is off IMO.
     
  2. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    72,578
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    A steep analog filter required for r2r 44.1kHz NOS will result in more phase shift than a gentler analog filter for OS 44.1kHz or DSD. This is because the a big part of the filter’s transition band will be in the audible range.

    This goes with music waveforms, not just test signals. There shouldn’t be any confusion with with frequency vs. time domain here because we are not looking at nor talking about FFTs here. We are talking time domain here.

    As far as the image, we only need need to treat the spatial aspect the same as the time aspect.

    I would like to reiterate a few points:
    1. Delta-sigma processing does not cause time domain fuckery.
    2. Oversampling was invented to avoid time domain and frequency domain fuckery in the audio band caused by brick-wall filters. The phase shifts caused by brick-wall filters were quite a big deal in the early CDP days.
    3. NOS R2R at 44.1 with an effective reconstruction (brick-wall) filter will have quite a bit of time domain and frequency domain fuckery in the audio band.
     
    • Like / Agree Like / Agree x 3
    • List
    Last edited: Jul 10, 2020
  3. Hammy

    Hammy Friend

    Friend
    Joined:
    Oct 3, 2015
    Likes Received:
    604
    Trophy Points:
    93
    Location:
    Oregon, USA
    My impression from this visual analog is that DSD looks like ass but shaped. Yet pleasing. The sonic implications being conveyed by this analogy are similar.
     
  4. Joe Bloggs

    Joe Bloggs Acquaintance

    Joined:
    Jan 14, 2017
    Likes Received:
    42
    Trophy Points:
    18
    Location:
    Hong Kong
    So... what about the Fourier transform?

    I find myself listening to test tones to suss out equipment deficiencies way more / easier / better than with listening to music. Is that just me?

    I also find the whole topic of this thread an unfortunate distraction from actual pursuit of good sound, but I guess that's also just me.
     
    • Like / Agree Like / Agree x 1
    • List
  5. k4rstar

    k4rstar Done his time

    Friend BWC
    Joined:
    Jun 11, 2016
    Likes Received:
    5,203
    Trophy Points:
    113
    Location:
    Toronto, Canada
    what about it? the Fourier transform is reductionist in nature. it does not help understand or interpret what I am interested in - which is why the acute listener can recognize subtle losses in music when listening to designs that are otherwise 'technically sound'. you can subscribe to a different approach than me, I won't argue just for the sake of arguing though.

    no one is forcing you to read it or post in it. the thread has over 16,000 views according to the forum. if my rambling caused even one person to reconsider their approach to DACs for their personal betterment, then mission accomplished.
     
    • Like / Agree Like / Agree x 3
    • List
  6. Joe Bloggs

    Joe Bloggs Acquaintance

    Joined:
    Jan 14, 2017
    Likes Received:
    42
    Trophy Points:
    18
    Location:
    Hong Kong
    Among other reasons why I subscribe to reductionism, it has been shown in actual controlled testing that actual people are much better at detecting all kinds of flaws in audio equipment when they are presented through test tones, rather than music. Here for just one example of poor results in the latter
    https://www.axiomaudio.com/blog/distortion/

    This suggests to me that taking the "holistic" approach may not be all it's cracked up to be...

    Edit: I've added test samples you can use to demonstrate this to yourself. I applied distortion to a 1kHz pure tone and then a piece of music everyone would be all too familiar with, with the same settings. Without resorting to opening them up in a DAW, can you tell which piece of music has been been distorted? Does the distortion sound anywhere near as apparent as in the pure tone?

    https://www.dropbox.com/s/ei7wougeek75ok6/sampleA.flac?dl=0
    https://www.dropbox.com/s/h2nqwi3msemii10/sampleB.flac?dl=0
    https://www.dropbox.com/s/61k0n8vd5p9ijvl/noteA.flac?dl=0
    https://www.dropbox.com/s/vffisgrrhp9k5yt/noteB.flac?dl=0

    I'm sorry I didn't notice that you were the actual OP, I wouldn't have posted my first post here if I did. But I guess we're kind of stuck with each other now ;)
     
    • Like / Agree Like / Agree x 1
    • List
    Last edited: Jul 11, 2020
  7. k4rstar

    k4rstar Done his time

    Friend BWC
    Joined:
    Jun 11, 2016
    Likes Received:
    5,203
    Trophy Points:
    113
    Location:
    Toronto, Canada
    we're going full hydrogenaudio here I see. isn't there a large fallacy in what you're saying? if the 'distortion' products can be discerned in a test tone and not in music then why should the listener or designer care about them? why the assumption that distortion products = bad and error correction schemes to remove distortion = good? these schemes have trade-offs at best and kill music at worst.

    all I know for sure is that long-term empirical testing led me to conclude that unmolested distortion products whether it be in-band phase shift in NOS dacs or the harmonic characteristics of a SET amp are less harmful to music than the attempts to remove/manipulate them for the sake of math and science. I am OK with being the flat Earth guy at the NASA convention
     
  8. Joe Bloggs

    Joe Bloggs Acquaintance

    Joined:
    Jan 14, 2017
    Likes Received:
    42
    Trophy Points:
    18
    Location:
    Hong Kong
    The distortion applied in the sample files is entirely of the push-pull odd-harmonic variety... but I digress. :oops:
     
  9. robot zombie

    robot zombie Friend

    Friend
    Joined:
    May 11, 2016
    Likes Received:
    1,456
    Trophy Points:
    93
    Location:
    Ennui, FL
    This has always sort of been an interesting debate. We talk a lot about distortion and how this translates to experience... a lot of philosophical and scientific interpretation of what can be heard and what we should be hearing.

    The thing is, it's like the friggin mind-body problem. The link to what we experience always seems to be missing, or at least... contentious.

    I can say for me, I have found myself leaning towards gear that is less 'neutral' or 'accurate'. Stuff I like tends to have some distortion. And it's not like I go looking for that. I tend to just look at the design and stated intent behind design choices (as much as they are inherently marketing, it's part of the intrigue,) before looking at what people are saying about it. And then maybe I try it. After that I'll look at some measurements. Stuff that measures closer to the current ideal, I swear I almost know how it will measure when I hear it. It's the sound of omission. While stuff I love, where it sounds like everything is there and everything is just right always seems to measure worse than normal in one way or another. And yet, it doesn't sound like a compromise... it's more like a revelation.

    I think on some level, I seek that out. Things that sound better than they have any business sounding. Not by price, value, or anything like that. But by virtue of what it is known to be.

    But of course that isn't always the case. There are minimum phase filters... any time I've had the option to use any kind of filter that cuts pre-ringing, I've hated it. Like, really and truly couldn't stand the sound.

    I think we try to make all of these connections between what we can understand and the experiences these things give rise to, but we don't actually understand the links that well. It's like there's something there in all of the stuff we uncover, but perhaps the connections aren't as they seem. Perhaps it's not even as much about distortion as we thought to begin with, you know? Or at least, it's a smaller piece of the picture than it's given credit for. It could just go hand in hand with other things that actually make a bigger difference. That would explain a lot for me. Because it seems like the further one goes down the rabbit hole of figuring out the characteristics of sound and how those pertain to the experience of sound, the more whatever you think you know gets turned upside down.

    I'm not a hardcore subjectivist. Ears can be quite unreliable, too. IME there is no real distinct trend I can point to in things that sound better to me. Which really complicates things. On one hand, senses aren't perfect at all. But I also think maybe there are big breakthroughs we just haven't had yet that'll make what we know now seem silly. So in that regard, it is really hard for me to settle on 'the one thing' that makes it all make sense. Because there's always something that by that standard, shouldn't sound better, but does.
     
    • Like / Agree Like / Agree x 2
    • List
    Last edited: Jul 11, 2020
  10. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    72,578
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Don't worry, lots of guys here even more flatter earth than you who reside the vinyl nut-job section. Why even worry about iterative math functions, minimum phase analog reconstruction filters, delta-sigma approximations, dithering/noise-shaping
    when they can be avoided.
     
    • Like / Agree Like / Agree x 1
    • List
  11. Josh83

    Josh83 Friend

    Friend
    Joined:
    Mar 1, 2017
    Likes Received:
    1,108
    Trophy Points:
    93
    Location:
    Charlottesville
    Yes, I still have a Solaris.

    I’ve always thought it had a slightly upper mid/lower treble-centric sound. But I wouldn’t say that it’s smoothed over. It doesn’t hit as hard as something like the Yggdrasil, but the Solaris is a fast DAC, to borrow a headphone term, IMHO. Transients are sharp and quick, but somewhat light feeling. It’s almost the Utopia of DACs in that sense.

    However, until your comment, I hadn’t really thought about electric guitars, specifically. Everything, to a certain extent, has less body on the Solaris than on something like the Yggdrasil. That definitely has a certain impact on distortion, among other things.

    But I really wanted to listen specifically to what you were talking about, so I pulled up some ‘90s favorites like Pearl Jam, Soundgarden, RATM, etc. What I noticed is that I could hear individual string articulation through distortion more on the Solaris than on other DACs. (I was comparing it to the Forssell, mostly, which has a very Yggdrasil tonality, except less deep bass.) Distorted guitars leaned towards the Tube Screamer sound than the Big Muff sound, for you guitarists, on the Solaris. However, it was really hard for me to say what the “correct” sound on most songs. (Small changes in mic position on the amp would create the same effect.)

    This is where the objective fundamentalists are wrong, of course. The Solaris is ruler flat. So’s the Forssell. Both have great linearity, etc. (Maybe the Solaris has lower jitter, because that’s Dave Hill’s forte, but that’s speciation.) Regardless, they sound like the FR is different. And I couldn’t tell you for sure if Mike McCreedy’s intro riff on “Brain of J” should sound like it does on the Solaris or the Forssell unless I heard the master tape. Everyone BSing about “true to the source” based on measurements alone is talking out their ass.
     
    • Like / Agree Like / Agree x 11
    • List
  12. skem

    skem Friend

    Friend
    Joined:
    Nov 1, 2017
    Likes Received:
    1,799
    Trophy Points:
    93
    Location:
    People’s Republic of Cambridge
    To @Josh83 ‘s point, the only way I try to evaluate DAC’s “correctness” is to listen to simply recorded and minimally processed acoustic-instrumental recordings. I find piano is especially challenging for DACs. It has transients, tons of overtones with different decay rates, and tremendous frequency coverage. Even so, unless you play piano you might not really notice.
     
    • Like / Agree Like / Agree x 3
    • List
  13. ChaChaRealSmooth

    ChaChaRealSmooth SBAF Gearmaster

    Staff Member Friend Gearmaster
    Joined:
    Dec 6, 2018
    Likes Received:
    6,943
    Trophy Points:
    93
    Location:
    Avalon
    Guess I'll throw my (new) two cents in here in regards to what the OP was trying to get at:

    I'll try to keep this mostly relegated to the DAC-part of my systems, but this is essentially my audio journey (and sometimes I'll have to talk about non-DAC things because it's relevant to my thought process). No, I don't expect anyone to really care to hear my "story," but f***ing whatever I'm posting it here anyhow.

    I started this hobby out connecting my phone directly to a Schiit Magni 3 and a Sennheiser HD 599. My phone, a S8 back then, obviously is far from an "audiophile" source, but that system actually brought me lots of enjoyment.

    Ironically, despite having bought a Schiit amp new, this is kind of where I began to go "anti-Schiit" at least in terms of DACs. Being new and not sure who to believe, I bought a Drop x Grace SDAC. This was an improvement over my phone, but it still wasn't the panacea that I thought it might be upgrading from a phone. Again ironically to my "anti-Schiit" buying habits, I later bought a B-stock Wyrd, which is the first time I realized how ass USB out of my PC really was. No joke, it was night and day difference. It was so stark my decidedly not-audiophile family who I convinced to listen to it told me there was a difference. This began the dispelling of me believing that measurements were everything*.

    *as far as why I thought this way in the beginning, I'm someone who comes from PC gaming. Almost everything about a computer's performance can be measured objectively, from frame-time consistency to actual FPS, time a CPU takes to render something in Blender, etc. Granted, computers have idiosyncrasies that cannot be explained just with numbers (which I'm sure many of you have experienced), but in general, I thought this applied to audio. Oh how wrong I was.

    From here, I thought of moving up, but this is when I had the side of me where I was the least certain who was actually "right" in terms of the extreme objectivists and subjectivists. This also coincided with when I joined SBAF (one of the best decisions of my audio journey; it's sappy but I truly appreciate this community). I was really close to buying a Schiit Gungnir A2, but I saw Marv say offhandedly somewhere about how the Matrix X-SPDIF 2 might actually be a good USB source (remember, I just realized how good the Wyrd was), so as a rando, I thought "heck, why not see if they'll send us one?"

    Well, interesting emails between me and Arthur (the USA Matrix rep) were exchanged where Marv was CC'd to make sure I didn't say anything too dumb, and somehow the X-SPDIF 2 turned into a X-Sabre Pro MQA, which was then sent to me first. I really thought it was fantastic (and it is), but trust me, I had no idea what to tell Marv when he asked me what I thought in PM. I was like "uhhh, I'm just a lowly rando?" One Matrix X-Sabre Pro loaner tour later though, I ended up buying the X-Sabre Pro because I got to hear it in my system and I liked it quite a bit**. It was the safe bet because at the time there wasn't a way for me to audition Gungnir A2.

    **the Matrix X-Sabre Pro does a lot right. Its timbre is actually quite good, and it sounds very modern, clean, and "hifi." In fact, it might be too clean and analytical for some; its attacks are quite sharp, its transients square instead of round, and while not to the extent of Convert-2, it doesn't really do nuance. That being said, I still think the X-Sabre Pro MQA absolutely deserves praise.

    Fast forward a couple months, I attended a meet at Marv's place, where I met @Donald North and tried out the then-unnamed Starlett. That amp absolutely blew me away and after further audition at my house, I bought the very first one made. It was the first time I heard something where I immediately knew upon first listen that it was special and I had to have it (keep in mind, this is my with the lenses of my preferences in mind; I don't necessarily recommend everyone who is looking for a TOTL tube amp to run out and buy one without due diligence). At the time, my system still had the X-Sabre Pro, and honestly, I was very content with my setup.

    Jumping forward yet again, @tommytakis kindly lent me his Bifrost 2. Those who've I've talked to will know that I thought the Bifrost 2 was really f***ing good, but ultimately something that I felt was too warm for my tastes. Specifically, the Schiit R2R timbre was something I really adored; the "tone density" just sounded very real and I loved the way it rendered my beloved piano (okay, there's tons of different kinds of pianos and not every recording uses a full-sized concert grand. Those things are gigantic and cumbersome even though they sound godly when tuned well. I take this into consideration while listening to classical piano/piano solos). Heck, instrumentation, vocals; I realized then the Bifrost 2's timbre was exactly what I wanted.

    There were problems though. I didn't feel the Bifrost 2 did as good of a job as X-Sabre Pro when it came to enabling me to hear the different layers in the music. An example was when there were harmonizing vocals; it was more difficult for me to pick out on Bifrost 2. This kind of musical interaction is something I really like listening to (don't ask me why), so ultimately the X-Sabre Pro stayed in my system. What I wanted, then, was something that was at least kind of like X-Sabre Pro in presentation, similar tonality, at least the same level of technicalities, but with the "tone dense" sound I got from the Bifrost 2. This led me to have slowly growing discontent with my own system.

    Ironically, the thing that made me snap and start seriously searching for a R2R DAC was the Sonnet Morpheus which I hated. Yes, I dislike the Morpheus, but I'll be damned if I don't admit that at least it had a kind of dense tone that I really liked. And the Morpheus isn't all bad; I can see why people like it. I don't personally, but I digress; that was the thing that made me realize that I wanted a change. I recognized that at the price I wanted to pay, I was most likely sidegrading, but that was okay with me.

    After many hmm's and hah's, I finally decided that what I wanted to buy was the Yggdrasil GS, and as I said in the relevant thread, it was everything I hoped for in a DAC and more. I liked it from when I first listened to it (similar feeling to when I first heard Starlett), and my appreciation for it only grew as it warmed up. It sounds grand, incisive, and lively, and while I hesitate to say "I love it" since I haven't had it for that long yet, I'm quite taken with it. And I finally have the tone dense timbre courtesy of the "reborn" R2R Schiit sound (as Marv would put it) which I wanted.

    So, I guess @k4rstar would say that I've now gone to a "righteous" path. Personally, I can see someone still preferring the X-Sabre Pro, but for me, the Yggdrasil GS is where it's at. I can say that my system as of right now (personal tastes changing notwithstanding) brings me that special feeling of "one more song" syndrome and actually makes it hard for me to critically listen because I'm enjoying myself that much. And I'm kicking myself for ever being swayed by both the measurements crowd and anti-Schiit crowd.

    Edit: note that I am not calling measurements useless. After many PM's with @atomicbob and having his help on my post on THD, I understand they can be useful tools. The problem obviously is the ASR fallacy thay lower noise floor/0.000000001% THD+N automatically means something is amazing. Hahaha....no. And don't get me started on Amir-bits.
     
    • Like / Agree Like / Agree x 28
    • List
    Last edited: Jul 14, 2020
  14. k4rstar

    k4rstar Done his time

    Friend BWC
    Joined:
    Jun 11, 2016
    Likes Received:
    5,203
    Trophy Points:
    113
    Location:
    Toronto, Canada
    @ChaChaRealSmooth awesome! you are now an R2R paladin, go forth and champion the righteous path

    [​IMG]

    @Josh83 thank you for your feedback. I will correlate what you refer to as lack of body as what ChaCha calls tone density. if equipment does not have this I typically do not want to listen, regardless of its other strengths

    @skem can you share some piano recordings you like, and use to evaluate. there are many different techniques used to record piano because of what you mention
     
    • Like / Agree Like / Agree x 1
    • List
  15. Melvillian

    Melvillian Friend

    Friend
    Joined:
    Jul 3, 2017
    Likes Received:
    1,376
    Trophy Points:
    93
    Location:
    Los Angeles
    I’ve noticed a lot of amplifiers have a difficult time with pianos as well. That’s another variable to consider
     
    • Like / Agree Like / Agree x 1
    • List
  16. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    72,578
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I'll try to change your mind with the Burl B2 Bomber DAC.
     
    • Like / Agree Like / Agree x 5
    • List
  17. mitochondrium

    mitochondrium Friend

    Friend
    Joined:
    Sep 1, 2017
    Likes Received:
    771
    Trophy Points:
    93
    Location:
    A Cell
    If you are willing to spend lots of money there might be a chance to get a solution which claims to give you multibit and DS at a high level in one piece of gear AMR DP-777 (SE):

    https://amr-audio.co.uk/products/dp-777-se/

    Using Philips TDA1541 for 44.1 kHz and a bastard 6bit /DS Chip for higher sample rates. There are tubes inside, even in the digital part.

    Why this particular approach? Explanation by the mastermind behind ifi and AMR Thorsten Loesch:

    https://www.audiostream.com/content/qa-thorsten-loesch-amrifi

    some more details about the comparison of the two formats:

    https://www.audiostream.com/content/qa-thorsten-loesch-amrifi-audiostream-addendum-pcm-vs-dsd

    This side of the pond you may fetch one used from time to time for under 2 k€.
     
    • Like / Agree Like / Agree x 1
    • List
    Last edited: Jul 14, 2020
  18. bozoc

    bozoc Rando

    Joined:
    Apr 27, 2017
    Likes Received:
    6
    Trophy Points:
    3
    Location:
    Croatia
    @k4rstar

    did you try any DIY DAC projects? plenty of those on diya

    I have generally found that designing a SD DACs that actually sounds good is not a trivial or cheap task.

    TDA1387 is peanuts these days, and what some implementations have managed to pull out of that chip is mind-blowing...
     
  19. k4rstar

    k4rstar Done his time

    Friend BWC
    Joined:
    Jun 11, 2016
    Likes Received:
    5,203
    Trophy Points:
    113
    Location:
    Toronto, Canada
    I have not yet - for experiments I am more keen on modifying early Philips-based machines instead. this will guarantee high quality, genuine vintage parts, already connected to an optical mechanism so we can easily test changes. the same cannot be said of many boards on eBay. even boards that look nearly identical can perform very different, or have fake replica TDA15xx, see here http://jelabs.blogspot.com/2020/04/the-good-bad-and-ugly-tda1543-nos-dacs.html

    with a Philips-based CD player you can start very simple, a NOS re-clock PCB ($30 on ebay) and replacing output op-amps. soon I would like to order some boards from Europe to try germanium transistor I/U stage, tube-based clock, tube-based S/PDIF buffer. these methods are already used in my DAC but I only understand the effect of their sum, not their individual contributions.

    [​IMG]

    it's not impossible to make an acceptable SD DAC, this is not my argument. it's just that designers today are focused on the wrong things. that is the measured noise performance using test tones. even those who pursue more audiophile marketed designs focus on the wrong things, femtoclocks, over-regulating every power supply section with LMxxx regulators, support for decoding the highest levels of DSD, fully differential output stages, and so on. this helps sell in the high-end market but it doesn't offer worthwhile performance gains over S/D DACs made over a decade ago or even cheap options on the market today.

    I could purchase some old professional sound card on ebay for little money, take the outputs directly from the DAC chip and connect it to a simple low-impedance buffer. it would require an old PC to work but it will already sound more natural than a modern high-end SD DAC.
     
    • Like / Agree Like / Agree x 3
    • List
  20. bozoc

    bozoc Rando

    Joined:
    Apr 27, 2017
    Likes Received:
    6
    Trophy Points:
    3
    Location:
    Croatia
    As far as the dreaded fake chip(including the TDA1387) stories go, I must say I have yet to see fake TDA1387, plenty of them were broken-probably from heat damage while being removed from old PC soundcards. I think Chinese engineers have better things to do than make TDA1387 fakes, lol.

    With any DAC, it is all about implementation, with R2R any output stage opamp that works in class AB tends to adversely affect the sound, and with TDA1387 even passive filtering works and it sounds wonderful!

    I would not say that with DS DACs focusing on femtoclocks, power supply section, support for decoding the highest levels of DSD, fully differential output stages, and so on is the wrong thing to do... DSs, unfortunately, need all that stuff to sounds decent IMO, properly implemented ofc among other things, meaning that with any good sounding DAC, a holistic approach is needed.
     
    • Like / Agree Like / Agree x 1
    • List

Share This Page